X-ray chest radiograph classification device based on improved dense connection network

Abstract

The invention relates to an X-ray chest radiography classification device based on an improved dense connection network. The device comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, and the processor is used for realizing the following method steps when executing the program: step 1, chest radiography preprocessing; 2, constructing and training a deep convolutional neural network; the method comprises the following steps of: modifying a classification network DenseNet121, adding an extrusion-excitation module into the network ina dense connection mode to form an improved dense connection network, constructing a loss function conforming to the task, and training the network by using a preprocessed chest radiography; and step3, testing the network and selecting an optimal network model: testing the network model obtained by each round of training on a test set, and selecting the network model with the highest average AUCvalue as a final model.

Description

technical field [0001] The invention belongs to the field of medical image recognition combining computer vision and medical images, and relates to a device for applying a deep learning algorithm to the discrimination of medical images and completing the classification of chest radiographs. Background technique [0002] Common lung diseases such as pulmonary nodules and pneumonia have a greater impact on the human body. A lung nodule is thought to be a possible precursor to cancer, and in patients with an underlying malignancy, it may also be evidence that the cancer has spread to the lungs. According to statistics, pneumonia is the largest cause of death among young children in developing countries and about 4 million people die from pneumonia every year in the world. In addition, abnormalities such as pleural effusion, emphysema, pneumothorax, atelectasis, and infiltration are also symptoms of some serious diseases. Early diagnosis and treatment of these diseases or abno...

AI Technical Summary

Problems solved by technology

[0004] Although the chest radiograph disease classification research based on deep learning has achieved good research results and progress, the following common problems still exist: (1) There is a problem of data imbalance in the X-ray chest radiograph dataset used for research , the sample size of some diseases is too small to train the network model, which brings difficulties to the recognition; (2) the network needs to learn the characteristics of 14 kinds of diseases at the same time, and it is difficult to take into account the accuracy of each kind of disease that leads to the network's recognition of different diseases. (3) The image features extracted by the network model over-learning lead to over-fitting problems and reduce the generalization performance of the model

Images